mersenne.ca

[James Heinrich]

Originally Posted by firejuggler

there is a slight bug in "tool to find P-1 done with the worst bounds." Probability calculator has a slight hiccup.

Not quite sure why those 3 exponents didn't get P-1 probability set in the database, but I confirmed it was just those three and updated them accordingly. Thanks for reporting.

[masser]

There's a problem in the calculator at the low end too:

[James Heinrich]

Originally Posted by masser

There's a problem in the calculator at the low end too:

Perhaps you could spell it out for me, nothing in your screenshot immediately jumps out at me as wrong.

[Viliam Furik]

Originally Posted by James Heinrich

Perhaps you could spell it out for me, nothing in your screenshot immediately jumps out at me as wrong.

I think he is referring to the exponent 14064697, which shows a wrong P-1 probability. It should be 5.637%, as shown in the exponent's page: M14064697

[storm5510]

Originally Posted by James Heinrich

Perhaps you could spell it out for me, nothing in your screenshot immediately jumps out at me as wrong.

The B1 for the first exponent is 1.5 million. The next one down, B1 is 155,000.

[James Heinrich]

Originally Posted by Viliam Furik

I think he is referring to the exponent 14064697, which shows a wrong P-1 probability

Thanks.

[firejuggler]

Another "0,000%" probability creepped up, 100,259,039

[James Heinrich]

Originally Posted by firejuggler

Another "0,000%" probability creepped up, 100,259,039

Still not sure how that happens, but I've added code to make sure if they do they'll be queued for updating and not show up in the list.

[petrw1]

Just sayin'

[De Wandelaar]

See attach GIMPS progress visualization : those indications appear simultaneously on the TF screen (a few minutes ago) :

• "Database contains data up to 2020-09-30 09:12:03"

• "the most recent data (2020-09-28)"

Anyway, data seem to be frozed at 2020-09-28 on the TF screen while they are ok when you choose the LL option.

[James Heinrich]

Originally Posted by De Wandelaar

Database contains data up to 2020-09-30 09:12:03
the most recent data (2020-09-28)

Both of these are true. There was some corrupt data that arrived in PrimeNet at 2020-09-30 09:12:03 (exactly how is under investigation, and no it's neither malicious nor a prime), at which point my nightly data import auto-aborted. After bypassing the corrupt entry I need to manually re-run the import, and of course today's data is 600% the average daily size (226k entries) so it takes a long time (~30m). If there are more corrupt entries I will need to bypass those as well and re-run the import again (hopefully not).

The visualization data cannot be updated until the nightly data import has completed, hence the delay there. I will make sure everything gets run and post back when things are up to date.